1.

CO2 Capture from SMR type flue gas using CESAR1 solvent at Technology Centre Mongstad (2022)

Sundus Akhter^a*, Ahmad Wakaa^a, Anette Knarvik^a,b, Erik Gjernes^c, Ida M. Bernhardsen^b, Muhammad I. Shah^a,c

aTechnology Centre Mongstad, 5954 Mongstad, Norway bEquinor ASA, PO Box 8500, 4035 Stavanger, Norway cGassnova SF, Dokkvegen 11, 3920 Porsgrunn, Norway

Abstract

The Technology Centre Mongstad (TCM DA) in Norway has investigated the CO₂ capture performance with the non-proprietary CESAR1 solvent for flue gases with CO₂ concentration like that of a SMR (steam methane reforming) furnace. The basis for this investigation is Equinor’s SMR unit at Tjeldbergodden methanol plant. Specific reboiler duty (SRD) is reported for 90% CO₂ capture from flue gases with 10 and 15 vol% (dry) CO2 content when using 12 and 18 m absorber packing height. Tests at 10 vol% (dry) CO₂ content confirmed that SRD levels below 4 GJ/ton CO₂ are achievable with only 12 m absorber packing height, but lower SRD values are achieved with 18 m absorber packing height. This illustrates that a more compact absorber will give lower CAPEX at the expense of a higher heat requirement for the CO₂ stripper. During the tests, precipitation was observed in the absorber packing despite that the overall absorber condition were unfavorable for this to occur. Also, foaming in stripper was observed and mitigated by the use of anti-foam.

1.  Introduction

The Technology Centre Mongstad (TCM DA), established in 2012, is one of the world’s largest facilities to test, verify and demonstrate different post combustion carbon dioxide capture (PCC) technologies. The company is a joint venture between Gassnova (the Norwegian state), Equinor, Shell and Total Energies with a common vision to facilitate development of carbon capture and storage (CCS) for the wide deployment of large-scale carbon capture technologies across industries. The facility is located next to the Equinor refinery in Mongstad providing two types of flue gas sources: (1) combined cycle gas turbine (CCGT) based heat and power plant flue gas (CHP) and (2) residual fluid catalytic cracker (RFCC) flue gas. The different flue gas sources enable TCM to mimic flue gases from different industries such as waste incineration, gas processing, steam reforming and oil refining.

Since the start-up in 2012, TCM has gained valuable knowledge and experience with the use of the non-proprietary solvents monoethanolamine (MEA) and CESAR1 for CO₂ capture. CESAR1 is a blend of 27 wt% 2-Amino-2- methylpropanol (AMP) and 13 wt% piperazine (PZ) and is considered to be a better solvent than MEA in terms of

thermal energy and stability [1,2]. CESAR1 was first tested at TCM as a part of the ALIGN-CCUS campaign in 2019 where the energy demand was found to be around 10% lower than that of 30 wt% MEA. Although the comparison is not straightforward, the tests provided valuable learnings with this solvent [3,4]. To further explore the solvent’s potential, the owners of TCM DA performed two campaigns with CESAR1 from April to December 2020. The goal of the campaigns was to carry out long term operation with this solvent, close knowledge gaps on the solvent performance with different flue gas compositions, to understand cost reduction potential of CO₂ capture with CESAR1 and perform thermal reclaiming [5] of the solvent.

This paper presents results from the part of the CESAR1 owner’s campaign which was carried out to understand the CESAR1 solvent’s potential for CO₂ capture from steam methane reformer (SMR) flue gas. Steam methane reforming is a key technology for converting natural gas to synthesis gas and is the dominant technology for hydrogen production. In this work, the CO₂ concentration was varied from 10 to 15 vol% (dry). The CO₂ concentration of 10 vol% mimic the SMR flue gas at the Equinor methanol plant at Tjeldbergodden [6]. The methanol plant at Tjeldbergodden is the largest in Europe. It produces approximately 900,000 tonnes of methanol per year and accounts for around 25% of total European methanol production. The methanol is produced using natural gas and pure oxygen as raw material. Natural gas is converted to synthesis gas by combined reforming (i.e. steam methane reforming followed by autothermal reforming) and the resulting synthesis gas is converted to methanol. Currently, the methanol plant at Tjeldbergodden emits around 300,000 tonnes CO2 per year. As most of this CO2 is generated in the SMR furnace by combustion of fuel, there is a considerable potential for CO₂ reduction. The fuel combusted in the SMR is a mixture of natural gas and purge gas from the synthesis gas, and the composition of the SMR flue gas will depend on the purge gas rate, and composition of the natural gas.

The first part of this paper presents the energy performance obtained with the CESAR1 solvent when varying the CO₂ concentration, absorber packing height and flow rates, while maintaining 90% CO₂ capture. The second part of the paper describes operational experiences with CESAR1 solvent at TCM.

2.  Assessment of CESAR1 performance with SMR flue gas

2.1 TCM amine plant description

Figure 1 shows a simplified illustration of the TCM amine plant in CHP mode. Flue gas from the blower is conditioned and saturated with water to the required temperature (normally 20-50 °C) in the direct contact cooler (DCC). CO₂ is captured in the absorber by the lean amine flowing downwards counter current with the flue gas. The depleted flue gas passes through two water wash stages before being emitted to the atmosphere. Rich amine from the absorber is pumped through the rich/lean cross plate heat exchanger to the stripper where the CO₂ is released using a steam-heated reboiler. The lean amine is then pumped back to the absorber.

The plant is flexible and offers the possibility to test several different configurations and a wide range of flue gas flow rates and compositions. Two strippers are available with capacities up to about 3,500 and 8,000 kg CO₂/h. The latter, named RFCC stripper, was in operation during the present test program. The lean amine solvent can be fed at various solvent flow rates utilizing 12, 18 or 24 m of the absorber structured packed bed (highlighted in yellow in Figure 1). The CO₂ content in the flue gas can be increased by recycling captured CO₂ to DCC inlet. A more detailed description of the amine plant is available elsewhere [7,8].

2.2 CESAR1 campaign mimicking SMR flue gas

The three phases A, B and C of the test program are presented in Table 1. The plant was operated to capture 90% CO₂ from a flue gas with 10 to 15 vol% (dry) CO₂, i.e. mimicking SMR flue gas. For each phase, the objective was to determine the optimal specific reboiler duty (SRD) by varying the solvent circulation rate.

The tests in Phase A were performed at 18 m absorber packing height to find the lowest possible SRD. In phase B, the packing height was reduced to 12 m. This was made to illustrate the consequence of utilizing a more compact absorber that reduces CAPEX at the expense of increased heat requirement and OPEX. During phase C, tests were conducted at 18 m and at a higher CO₂ concentration to better understand the impact of flue gas CO₂ concentration at a target capture rate of 90%. All tests were made with CHP flue gas including CO₂ recycling and with the RFCC stripper in operation. CO₂ recycling increased the CO₂ content in the flue gas into the absorber from about 4 to 10 and 15 vol% (dry). The target CESAR1 solvent concentration during the campaign was 40 wt% comprising 27 wt% AMP and 13 wt% Piperazine. This concentration was similar as used in the previous ALIGN-CCUS campaign at TCM [4], however, variations occurred both in total amine concentration and Piperazine and AMP ratio. Calibration and validation of analysers were carried out before and during the campaign to ensure good data quality on solvent performance and compliance with emission permit.

Table 1 Operational parameters for tests mimicking SMR flue gas.

¹Captured CO₂ for test A5, B7* and C7.

The amount of CO₂ being captured can be calculated based on either the mass balance over the absorber; i.e. CO₂ in versus CO₂ out of the absorber or on the CO₂ mass flow out of the stripper. The results presented here are based on the latter. The CO₂ capture rate is then calculated as mass flow of captured CO₂ versus the CO₂ mass flow into the absorber. SRD is calculated based on the steam side enthalpy difference over the reboiler heat exchanger divided by the amount of captured CO₂. TCM is well equipped with multiple analysers and flow meters for each of the three gas flows. The present analysis is made using the same selection of analysers and flow meters as [9].

Each test was operated up to 24 hours targeting stable operation and each of the data points presented in Figure 2 to Figure 4 represent an average over typical 2 hours of stable operation within that test interval. The data quality may be affected by steam quality and moisture in the CO₂ product flow out of the stripper. For phase A and part of phase B, steam temperature into the reboiler heat exchanger was close to saturation temperature and thus introduced some uncertainty on the steam quality. It should be noted that for phase A, B and C, the measured moisture content out of the stripper fluctuated rather lively.

2.3 Results from flue gas with 10 vol% CO₂

The energy required to capture 90% of the CO₂ from the CHP flue gas with recycling to mimic 10 vol% SMR

furnace type of flue gas was determined during phase A and B tests.

Figure 2 (a) shows SRD at 90% capture rate for phase A tests utilising 18 m packing height. During the tests the flue gas flow rate was kept constant at 48,000 Sm³/h, while the solvent flow rate was varied. The optimum steam consumption achieved for tests A5 and A1 were approximatively the same, i.e. 3.2 GJ/ton CO₂ at lowest L/G ratio around 1.9 kg/Sm³, at lean solvent flow rates 90,800 kg/h and 92,100 kg/h, respectively. Note that A1 test was conducted at 35 °C while A5 tests was carried out at 37 °C. The increase in temperature was applied to control precipitation in the absorber. As experienced from previous campaigns [4], increasing the flue gas inlet temperature has a negative effect on the SRD. However, Figure 2 (a) shows that SRD during A5 test is slightly lower than A1 test. This may be due to instability in the plant operation during A1 test and since calibration was not performed for the IR-high instrument at absorber inlet. The IR-high instrument showed higher values while the actual CO₂ concentration during A1 test was 9.5 vol% (dry). Calibration was made after the A1 test and point A5 is considered as the optimum for the phase A tests. After the B series the A5 test was repeated with an SRD of 3.3 GJ/ton CO₂. This illustrates reduced reproducibility due to minor changes in operational conditions etc.

Phase B utilised 12 m absorber packing height while maintaining capture rate at 90% as well as flue gas flow at 48,000 Sm³/h. SRD plotted against L/G ratio for tests from phase B is shown in Figure 2 (b). During the B1-1 and B2 tests, there was an issue with controlling the steam temperature and pressure to the RFCC reboiler resulting in lower CO₂ capture rate. In addition, precipitation was observed and there were frequent foaming issues in the stripper packing which led to a poor stripper performance. Several measures were taken such as increasing the flue gas inlet temperature from 37 °C to 38 °C, to be within the precipitation free zone, or even flushing the absorber sometimes to remove precipitation, and addition of antifoam to reverse the foaming in the stripper. Since the initial results of phase B tests were not satisfactory and the variation in the SRD values was small ranging from 4.0 to 4.3 GJ/ton CO₂, phase B repeat tests were carried out including two new tests B6* and B7*. B1 in Figure 2 has a low SRD but the test was at only 85% capture rate. When considering B1 (repeat), B2 (repeat), B4 (repeat), B5 (repeat), B6* and B7* tests, the SRD values were in the range 3.8 – 4.2 GJ/ton CO₂. The minimum energy penalty achieved during phase B repeat was

3.8 GJ/ton CO₂ at 113,300 kg/h lean solvent circulation rate and at L/G ratio around 2.4 kg/Sm³ during B7* test. This test series has also two more test points confirming that SRD below 4 GJ/ton CO₂ is achievable utilising 12 m absorber packing.

2.4 Comparison of absorber packing height

Using results from phase A and phase B, the effect of packing height on energy performance when operating with SMR type flue gas (10 vol% dry CO₂) can be assessed as shown in Figure 3. Reducing the packing height from 18 to 12 m gives an increase in SRD when comparing the optimum test points, i.e. from 18 m series A5 with an SRD 3.2 GJ/ton CO₂ to the 12 m series B7* with an SRD 3.8 GJ/ton CO₂. There is also a corresponding increase in the optimum L/G ratio from 1.9 to 2.4 kg/Sm³. First of all, this illustrates that selecting a more compact absorber that can reduce CAPEX will have a consequence on the OPEX through increased heat requirement for the CO₂ stripper as well as increased pumping power for solvent circulation. Thus, a shorter absorption column may be beneficial for installations with limited space available or if low-cost thermal energy is available.

1. Results from flue gas with 15 vol% CO₂

2.5Results from flue gas with 15 vol% CO₂

During phase C the CO₂ level in the flue gas into the absorber was increased to 15 vol% (dry) and in order to be within the design of the stripper, the flue gas flow is decreased to 34,000 Sm³/h. SRD plotted against L/G ratio for tests from phase C at 18 m packing height and 90% capture rate is shown in Figure 4. The SRD values were in the range 3.2- 3.5 GJ/ton CO₂. The minimum energy penalty achieved during phase C was 3.2 GJ/ton CO₂ at 98,400 kg/h lean solvent circulation rate and at L/G ratio around 2.9 kg/Sm³ during C7 test. The SRD level is similar to test A5 above, however, lower L/G ratios were not explored due to time constraints so the optimal case could be even further to the left of the lowest SRD value achieved from C7 test.

3.  Operational experience

3.1 Precipitation in the lower absorber packing and mitigating actions

According to TCM’s laboratory tests prior to ALIGN-CCUS campaign, CESAR1 solvent should not precipitate at planned absorber operation conditions [4]. However, at TCM precipitation has turned out to be a reoccurring issue with the CESAR1 solvent. The TCM absorber design is with rectangular cross-section and both temperature and concentrations may vary and create local conditions for CESAR1 to precipitate. Already during the ALIGN-CCUS campaign TCM gained knowledge under which conditions precipitation can occur, how to handle and dissolve precipitates.

Precipitation was most easily detected by increase in the pressure drop in the lower absorber packing. If larger amounts of solvent precipitate, it can also be noticed by significant decrease in the solvent concentration. At stable operation, without precipitation, the lower packing pressure drop is stable (+/- 0.1 mbar), and variations are mainly caused by changes in lean solvent or flue gas flow rates.

With precipitation two pressure drop profiles in absorber lower bed are observed:

1. Continuous fast increase of the pressure drop (100% increase in few hours).
   1. Unstable and increasing or decreasing pressure drop (+/- 1 to 2 mbar) at stable absorber operation or with slight variations of operation parameters.

During normal operation without precipitation, the four temperature sensors at the same elevation at the bottom of the lower absorber packing usually indicate a deviation of +/- 5 °C, and the temperature close to the absorber wall is higher than the other locations (unless no reaction is occurring). This deviation is most likely caused by higher local CO₂ loading and may indicate a poor distribution close to the walls in normal operation.

With precipitation the temperature sensors often indicate similar temperatures (+/- 1 °C in some cases). However, with high precipitation (pressure drop profile 1), flooding may occur on the whole packing area, causing a uniform temperature of the packing. Significant modification of the operation parameters or absorber flushing is required in order to control the pressure drop. In case of pressure drop profiles 2, an increase or decrease of the pressure drop is also observed at several tests at almost identical parameters, despite of better conditions for avoiding precipitation. This is likely caused by the local accumulation of precipitation close to the wall as the concrete absorber walls are not insulated and poor gas distribution is expected at the corners and walls.

During the start of phase A at 18 m absorber packing height, 10 vol% (dry) CO2 concentration and 90% CO2 capture rate, there was a slight increase in the differential pressure in the lower absorber bed which indicated that precipitation might be occurring. This was dealt with by increasing the flue gas inlet temperature from 35 °C to 37 °C in attempt to avoid precipitation. Although a higher flue gas inlet temperature negatively impact the steam consumption, it was prioritized to run tests at higher temperatures in order to avoid precipitation. However, towards the end of phase A tests and despite the increased temperature, the pressure drop across the lower absorber packing was increased drastically from 7 to 14 mbar indicating a significant precipitation event. An absorber flushing was required to dissolve the precipitate. This was carried out by turning off the flue gas feed blower and increasing both the lean solvent flow rate and temperature. As a result of the flush test, the differential pressure in the lower absorber bed was reduced to 7 mbar indicating that the precipitate had been cleared. The measured and the calculated pressure drop during phase A testing are shown in Figure 5 (a).

During phase C tests at 18 m packing height, 15 vol% CO2 (dry) in the flue gas and 90% CO2 capture rate, the increasing differential pressure in the lower absorber packing was a recurring issue even if maintaining a high flue gas temperature of 40 °C. Therefore, two tests were dedicated to investigating further the differential pressure rise, see Figure 5 (b). The first test was run to provoke the differential pressure to be increased. During this test, the pressure drop increased from 4 mbar to 7 mbar. In the second test the water which is usually dumped to the absorber sump from the lower water wash was led to the 3^rd water wash (the unused upper absorber bed) instead to see if this could affect positively on the differential pressure in the lower absorber bed (i.e. by contributing to diluting and flushing). However, the effect on the pressure drop was limited, and the differential pressure stabilized at 5.9 mbar as shown in Figure 5 (b).

3.2 Foaming evidence in the RFCC stripper during tests mimicking SMR flue gas

Analyses of the stripper temperature profiles at TCM revealed that CESAR1 solvent tends to foam, and this depends on several factors that may vary in the plant operation. At times stripper profiles show that foaming may start some hours after plant start-up, even with a fresh solvent and may be increased or decreased. Operation parameters that are suspected to cause and/or influence the formation of foam are mostly not predictable or repeatable. Among these, the high stripper temperature and consequently high evaporation rate of solvent at the bottom of the stripper packing. Other potential reasons for formation of foam could be impurities in flue gas, solvent properties, particles of solids and corrosion, and design of stripper column internals.

The main consequence of foaming observed is unstable stripper operation. Hence, foaming leads to decreased stripper efficiency and thereby not allowing to achieve optimal test conditions. As foaming has a direct impact on the regenerator performance, more steam will be needed to regenerate the solvent and achieve the target CO₂ capture rate. Hence, the specific reboiler duty SRD values can increase by 0.1 GJ/ton CO₂ up to 0.25 GJ/ton CO₂ in the worst cases of foaming.

Foaming in the RFCC stripper was an intermittent issue throughout the test program particularly during original phase B tests at 12 m absorber packing height (see section 2.3). As the results from these tests were not promising due to several periods of high foaming or flow maldistribution, phase B tests were repeated to reverse the foaming issue by injecting a small amount of antifoam. Although foaming during phase B repeat and phase C was limited, a 2 liter of antifoam addition was able to remove the strong foaming tendency and led to lower steam requirements in the stripper reboiler and consequently lower energy numbers (SRD). However, the duration of antifoam effect is variable and could be limited to only a few hours.

Based on literature [10] excessive antifoam dosing is not recommended as its function can be reversed and be a foam promoter. Future CESAR1 tests at TCM may confirm the minimum injection rate and if lower injection rates at short intervals reduce the risk of excessive antifoam dosing. It is also recommended to minimize risks of local high gas flow or liquid flow in the stripper packing by ensuring a suitable vapor flow distribution from the stripper reboiler to the stripper packing.

Future CESAR1 tests might also consider testing the solvent at relevant temperature, pressure and packing with representative flue gas to be able to confirm the eventual impact of specific decomposition products to foaming. If foaming is caused by impurities or degradation products, then it could be possible that activated carbon bed operation and/or thermal reclaimer unit can help to reduce foaming tendency.

Conclusion

The three test series (phase A, B and C) discussed above with CESAR1 solvent showed interesting results for 90% CO₂ capture from flue gases with comparable levels of CO₂ to that of a SMR furnace. At 10 vol% (dry) CO₂ the minimum SRD was found to be 3.2 GJ/ton CO₂ when utilizing 18 m absorber packing. For the same flue gas conditions

but with absorber packing height reduced to 12 m, the minimum SRD became 3.8 GJ/ton CO₂. This test series has two more test points confirming that SRD below 4 GJ/ton CO₂ is achievable with only 12 m absorber packing. The 18 and 12 m cases with 10 vol% CO₂ illustrate the consequence of utilizing a more compact absorber that reduces capex at the expense of increased heat requirement for the CO₂ stripper. The third test series was operated at 18 m absorber packing with CO₂ in flue gas increased to 15 vol% CO₂ (dry). Similar SRD level as the corresponding 10 vol% CO₂ case was found. However, the minimum SRD was likely not demonstrated since the point where SRD stops decreasing when decreasing L/G had not been reached.

During the test program there was precipitation in the lower absorber packing in spite of that the conditions in the absorber should not be favorable for this to occur. Precipitation was mitigated by increasing the flue gas inlet temperature and flushing of the absorber was also required. The cause may be related to the absorber design and local zones with poor gas distribution at lower temperature close to the absorber walls as well as corners. Foaming was observed in the stripper and introduced unstable stripper performance and less efficient stripping. Injection of anti- foam reduced the foaming tendency, but the effect was temporary. Better understanding of the onset of precipitation and foaming should be included in future CESAR1 tests at TCM.

The basis for this investigation is Equinor’s SMR unit at Tjeldbergodden methanol plant which is the largest methanol plant in Europe. The test campaign provides value to the industry as it helps to reduce both technical and financial uncertainty associated with CO₂ capture. The campaign results give an indication on what SRD values one might expect when operating a CO₂ capture plant.

Acknowledgments

The authors gratefully acknowledge the staff of TCM DA, Gassnova, Equinor, Shell and TotalEnergies for their contribution and work at the TCM DA facility. The authors also gratefully acknowledge Gassnova, Equinor, Shell, and TotalEnergies as the owners of TCM DA for their financial support and contributions.

References

• Further Assessment of Emerging CO₂ Capture Technologies for the Power Sector and their Potential to Reduce Costs: IEAGHG Technical Report, 09. 2019.
• G. Wiechers, P. Moser, C. Benquet, J. Gibbons, J. Monteiro, A. Hartono, K-. A. Solli, H. Knuutila, Guidelines for effective solvent management, Deliverable Nr. D1.2.6, ALIGN-CCUS.
• Accelerating Low carbon Industrial Growth through CCUS: ALIGN-CCUS, project homepage https://www.alignccus.eu/, website accessed: 15.11.2021.
• C. Benquet, A. Knarvik, E. Gjernes, O. A. Hvidsten, E. R. Kleppe, S. Akhter, First Process Results and Operational Experience with CESAR1 Solvent at TCM with High Capture Rates (ALIGN-CCUS Project), 15th International Conference on Greenhouse Gas Control Technologies,GHGT-15, 15th-18th March 2021 Abu Dhabi, UAE. http://dx.doi.org/10.2139/ssrn.3814712

• M.Campbell, S. Akhter, A. Knarvik, M. Zeeshan, A. Wakaa, CESAR1 solvent degradation and thermal reclaiming results from TCM testing. To be presented at the GHGT-16 Conference: 23 – 27 October 2022, Lyon, France.
• Equinor onshore facilities, Tjeldbergodden industrial facility, https://www.equinor.com/energy/onshore-facilities, website accessed: 26.08.2022.
• E. Gjernes, S. Pedersen, T. Cents, G. Watson, B.F. Fostås, M.I. Shah, G. Lombardo, C. Desvignes, N.E. Flø, A.K. Morken, T. de Cazenove,L. Faramarzi, E.S. Hamborg, Results from 30 wt% MEA performance testing at the CO2 Technology Centre Mongstad, Energy Procedia, 114, 2017, 1146-1157, https://doi.org/10.1016/j.egypro.2017.03.1276

• M.I. Shah, G. Lombardo, B.F. Fostås, C. Benquet, A.K. Morken, T. de Cazenove, CO₂ Capture from RFCC Flue Gas with 30w% MEA at Technology Centre Mongstad, Process Optimization and Performance Comparison, 14th Greenhouse Gas Control Technologies Conference Melbourne 21-26 October 2018 (GHGT-14), https://ssrn.com/abstract=3366149
• M.I. Shah, E. Silva, E. Gjernes, K.I. Åsen, Cost Reduction Study for MEA based CCGT Post-Combustion CO₂ Capture at Technology Center Mongstad. Proceedings of the 15th Greenhouse Gas Control Technologies Conference 15-18 March 2021, Abu Dhabi, UAE. http://dx.doi.org/10.2139/ssrn.3821061
• [10]A.M. Al-Dhafeeri, H.A. Nasr-El-Din, Characteristics of high-permeability zones using core analysis, and production logging data. J Pet Sci Eng – J PET SCI Eng 2007;55:18–36. https://doi.org/10.1016/j.petrol.2006.04.019.

Scale-up testing of advanced Polaris™ membrane CO2 capture
technology (2022)

Vincent Batoon^a, Adam Borsaly^a, Carlos Casillas^a, Thomas Hofmann^a, Ivy Huang^a,
Jay Kniep^a, Tim Merkel^a,*, Craig Paulaha^a, Witopo Salim^a, Erik Westling^a
aMembrane Technology and Research Inc., 39630 Eureka Drive, Newark CA 94560, United States

Abstract

Membrane processes offer some advantages when applied to post-combustion CO₂ capture, including no hazardous chemical storage, handling or emissions issues, simple passive operation, tolerance to high SOx and NOx content, recovery of flue gas water, and use of only electric power. Membrane Technology and Research, Inc. (MTR) of Newark, California, USA has been working on the development of membrane capture technology for more than a decade. Along the way, key material and process innovations have been made, including:

1. A new class of high permeance membranes, called Polaris™, was developed. This membrane is tenfold more permeable than prior commercial CO₂ membranes, resulting in a large decrease in membrane area and capital cost.
2. A membrane selective recycle process was developed. The patented process uses combustion air as a sweep stream to generate driving force for transmembrane CO₂ transport. This selective recycle approach is particularly useful for achieving high capture rates.

Recently, in an effort to optimize capture system footprint while minimizing parasitic pressure drops, MTR developed new planar module stacks designed specifically for low-pressure flue gas operation. These new modules feature the ability for fine control of the flow path on both sides of the membrane, allowing optimal performance. In prior work, these improvements were validated in prototype testing on a small pilot system at the U.S. National Carbon Capture Center (NCCC). At the same flow rate, the new planar module had about 70% lower pressure-drop compared to conventional modules, which would yield savings of approximately 10 MWe in fan power on a full-scale system. In addition to energy savings, we believe the regular geometry of the new module design is more amenable to automated fabrication methods that will reduce cost.

In this paper, the performance of these Polaris modules during an engineering-scale field test at the Technology Centre Mongstad (TCM) in Norway will be reviewed. TCM is a joint venture between Gassnova (on behalf of the Norwegian state), Equinor, Shell, and TotalEnergies. The membrane system at TCM treats a slipstream of flue gas from the Equinor residual fluid catalytic cracker (RFCC) unit. The membrane system includes both the primary capture step and the selective recycle step of the MTR selective recycle process. Depending on system operating parameters, between 50 – 90% of the CO₂ in the inlet slipstream can be captured. Figure 1 shows a conceptual drawing of the membrane module stacks in a containerized form and a picture of an actual one of these containers on the MTR small pilot system at TCM. Future, larger capture systems will simply use multiples of this modular building block.

Work began in 2019 to develop a new site at TCM for hosting 2^nd generation CO₂ capture technologies. Utility hook-ups and site preparation work was completed at this site prior to the arrival of the MTR test system at TCM in late May 2021. TCM provided site support and coordinated with MTR on installation, shake-down operations, and hot commissioning activities. The MTR test system was commissioned on flue gas in July 2021 and continued operation through February 2022. The goals of the field test were to identify optimum conditions for different CO₂ capture rates (50 – 90%), to validate the low pressure drops of the planar modules, and to determine system performance under inlet CO₂ concentrations between 14 and 25% corresponding to different industrial cases. CO₂ capture rates up to 91% were achieved during the field test and the planar module pressure drops were even lower than expected (2 kPa vs. 10 kPa). These results from the TCM field test will be used to further refine the MTR CO₂ capture process for power plants and other large point sources, such as cement or steel plants.

1.  Introduction

1.1 Background

Over the past decade, the U.S. Department of Energy (DOE) has funded a large research effort to identify low- cost ways to capture CO₂ from the emissions of large point sources, such as power generation facilities, to mitigate the climate impact of unabated CO₂ emissions. Coal-fired power plants have been a particular focus for CO₂ capture efforts because of the large installed base of these plants, which produce almost 40% of U.S. CO₂ emissions. Moreover, the relative low cost and large domestic supply of coal suggests that this fuel will remain important to power production for the foreseeable future [1, 2].

Currently, amine absorption is the leading candidate technology for post-combustion CO₂ capture. This capture approach is a proven technology used successfully to remove CO₂ from industrial gas streams for decades. Initial capture systems at commercial power stations such as Boundary Dam in Canada and WA Parish in Texas use amine absorption. However, studies indicate that amine absorption, when applied to flue gas CO₂ capture, is costly and energy intensive [2,3]. In addition, amine systems have environmental issues related to emissions and handling of hazardous chemicals. As a result, DOE is funding development of transformative new technologies based on advanced solvents, membranes, or hybrids with a goal of reducing capture costs to $30/tonne or less.

Among advanced capture technologies being considered are a number of membrane approaches [4-7]. Membrane processes offer some advantages when applied to post-combustion CO₂ capture, including no hazardous chemical storage, handling or emissions issues, simple passive operation, tolerance to high SO_x and NO_x content, recovery of flue gas water, and use of only electric power, which could be from low CO₂-footprint sources (renewables). The

main challenge for post-combustion capture membranes is the low partial pressure of CO₂ in flue gas, which results in very large membrane area being required because of the small driving force for separation. Some years ago, working with DOE, MTR made two transformative innovations to address this problem:

• A new class of high permeance membranes, called Polaris™, was developed. This membrane was approximately tenfold more permeable than prior commercial membranes, resulting in a large decrease in required membrane area, and thus capital cost.

• A membrane selective recycle process was developed. This patented process uses combustion air as a sweep stream to generate driving force for transmembrane CO₂ transport [8]. The separated CO₂ is recycled to the boiler with air. This design increases the concentration of CO₂ in flue gas, which reduces the energy and capital required for capture.

Subsequently, MTR has worked with DOE to develop these innovations into a cost-effective CO₂ capture process. This effort has included the first test of membrane modules with coal-fired flue gas at the Arizona Public Services (APS) Cholla plant in 2010; the accumulation of >11,000 hours of flue gas operation for Polaris modules on a bench- scale 1 tonne/day (TPD) system at the National Carbon Capture Center (NCCC); scale-up of first generation Polaris to a 20 TPD small pilot system, and successful operation of this system on a flue gas slipstream at NCCC and in integrated boiler testing at B&W.

1.2 Polaris membrane

Several years ago, MTR developed a class of composite membranes called Polaris where the selective layer is based on polar polymers that are extremely permeable to CO₂ and other polar species. This first generation (Gen-1) Polaris membrane set the standard against which all post-combustion capture membranes are now compared. With an average CO₂ permeance of 1,000 gpu and a CO₂/N₂ pure-gas selectivity of 50, Gen-1 Polaris was a step-change improvement over typical commercial CO₂-selective membranes used for natural gas treatment (which offer a CO₂ permeance of around 100 gpu combined with a pure-gas CO2/N2 selectivity of 30). This improvement is illustrated in Figure 2, where membrane performance is compared in the form of a trade-off plot of CO₂/N₂ selectivity versus CO₂ permeance. Better membranes will have properties that move up and to the right on this plot.

In addition to showcasing the benefits of Polaris over conventional membranes, Figure 2 also shows some of the more recent improvements in the performance of Polaris membranes. A second generation (Gen-2) Polaris membrane has been scaled up to pilot production. This membrane offers about double the CO₂ permeance with similar selectivity to the base case Polaris. These Gen-2 membranes have been fabricated into prototype modules, and validated in bench-scale testing at NCCC. Recently, even higher permeance Gen-3 Polaris membranes (3,000 gpu) have been produced at the lab scale. These improvements are important because the size and capital cost of a membrane unit scales almost linearly with membrane CO₂ permeance. Thus, these new Polaris membranes would yield a system with one-half to one-third as many membrane vessels as the Gen-1 membranes; this would be a dramatic reduction in system size and cost.

1.3 MTR post-combustion CO₂ capture

In addition to a membrane with good separation performance, an energy-efficient and affordable process design is required to make membranes competitive for post-combustion CO₂ capture. Prior membrane process studies have produced the following general conclusions about using membranes for post-combustion capture [4, 8, 9]:

• To capture CO₂ from flue gas, a membrane process needs partial pressure driving force. This driving force can be generated by either (a) compression on the feed-side or (b) a vacuum on the permeate-side of the membrane. The energy required is considerably lower for a vacuum process because the vacuum only has to pump the flue gas that permeates the membrane (about 10% of the total flue gas, and largely CO₂), whereas a feed compressor has to pressurize all of the flue gas (CO₂ plus the bulk N₂). While a vacuum process uses less energy, it requires a larger membrane area, because the CO₂ partial pressure difference across the membrane is small. Consequently, an energy-efficient vacuum-driven process requires very permeable membranes.

• In addition to large membrane area or power requirements, single-stage membrane designs are unable to produce high-purity CO₂ combined with high CO2 capture rates. In fact, a single-stage membrane process alone cannot produce high-purity CO₂ in the permeate with 90% CO₂ capture, regardless of the membrane selectivity. This is because the system performance is limited by the pressure ratio across the membrane – that is, the ratio of the feed pressure to the permeate pressure. Higher pressure ratios for flue gas treatment could be generated, but at a high energy and capital cost. With a maximum affordable pressure ratio of ~10, the ideal membrane selectivity for flue gas CO₂ capture is about three to five times the pressure ratio, or a CO₂/N₂ selectivity of 30-50 [10]. Beyond this point, it is much more important to increase membrane permeance to reduce area requirements rather than trying to improve selectivity.

To overcome these driving force issues and achieve a relatively high CO₂ capture rate and high purity, membrane developers have proposed multi-step/stage membrane designs and/or hybrids with other separation technologies (cryogenics, adsorption, etc.). For example, MTR developed the selective recycle process design shown in Figure 3 [8]. This process uses a permeate vacuum in a first membrane step to efficiently generate a pressure ratio that will lead to capture of about 70% of the inlet flue gas CO₂. The partially-treated flue gas leaving this primary CO₂ removal unit is then sent to a second membrane step that utilizes a sweep gas of combustion air to selectively recycle CO₂ to the boiler and drive the overall CO₂ recovery of 90% or higher.

This CO₂ recycle design has a number of features that optimize membrane system performance:

• Because it is a two-step membrane design, all of the flue gas CO₂ does not have to be removed in a single membrane step. This allows the first-step membrane to operate efficiently as a partial capture step (~70%) with a relatively high partial pressure of CO₂ on the feed-side.

• The second membrane step performs the difficult task of removing CO₂ to very low levels (i.e., to reach >90% capture). This step uses an air sweep stream to maintain separation driving force by keeping a relatively low partial pressure of CO₂ on the permeate-side. Because the air stream is already being blown into the boiler as the oxidant for combustion, this sweep gas provides an essentially free separation (i.e., no compressors or vacuum pumps are used in this step).

• The concentration of CO₂ in the flue gas leaving the boiler is increased (for example, from 12% to 18% CO2) because CO₂ is recycled to the boiler with the air sweep stream. This enrichment makes CO₂ capture in the first membrane step easier due to the higher CO₂ partial pressure.

1.4 Membrane module design

One of the key issues for a membrane post-combustion capture system is how to balance the desire for a small system footprint with the need to process large volumetric flows and minimize parasitic pressure drops. The pressure drop issue is particularly important because for a full-scale (550 MW_e power plant) membrane capture system, each 1 psi of pressure drop through the membrane unit amounts to approximately 3 MW_e of required blower energy. In previous techno-economic analysis (TEA) studies of a commercially mature n^th-of-a-kind (NOAK) full-scale MTR membrane post-combustion capture system, MTR assumed 1.5 psi (~10 kPa) pressure drop through each of the membrane units shown in Figure 4. The ability to reach this pressure drop target while maintaining a compact system size and good membrane performance depends on the module design.

In earlier work, MTR addressed these module design issues by first adapting existing spiral-wound module technology for low-pressure operation. Early spiral modules showed unacceptably high pressure drops. Through changes in spacer geometry and module configuration, the best spiral-wound sweep modules were able to reach a sweep-side pressure drop of just under 4 psi. Later, recognizing some of the limitations associated with tortuous flow in spirals, new plate-and-frame modules were designed specifically for flue gas operation. The most important feature of these new modules is the ability for fine control of the flow path on both the feed and sweep sides of the membrane, which can be used to minimize pressure drop. Under equivalent laboratory conditions, plate-and-frame modules with similar packing density to spiral-wound modules can achieve a pressure drop of less than 0.5 psi.

To validate these laboratory improvements, a prototype planar module was built and tested in a side-by-side comparison with spirals on the small pilot system at NCCC. Figure 4 shows these results, which confirm the improved performance of the new modules. At the same flowrate, the plate-and-frame module had about 3 psi lower pressure drop compared to the sweep spirals. Scaled to a full power plant, this would yield savings of approximately 10 MW_e in fan power. In addition to energy savings, we believe the regular geometry of the new design module is more amenable to automated fabrication methods that will reduce cost.

2.  Previous small pilot field tests

2.1 NCCC slipstream field test

Using field test performance data and experience from bench-scale field tests at APS and NCCC, MTR designed and built a small pilot test system to treat 20 TPD from a Plant Gaston coal-fired flue gas slipstream at NCCC (equivalent to 1 MW_e of coal-fired power generation). This small pilot membrane system exhibited stable performance of the Gen-1 Polaris membrane over a six-month period in 2015, operating in both winter and summer conditions. During this field test, the system averaged 87% CO₂ capture. Figure 5(a) shows a picture of the MTR small pilot system at NCCC. Also shown in Figure 5(b) for comparison are the 10 TPD pilot solvent test unit (PTSU) and a 20 TPD advanced amine system at NCCC. This picture illustrates the compact nature of the MTR membrane system compared to other separation technologies.

The small pilot test system at NCCC was also utilized to validate improvements in membrane module design. A prototype planar module vessel was built and installed at site for a side-by-side comparison with spiral-wound modules under the same field test conditions. As previously shown in Figure 4, the vastly improved pressure drop performance of the plate-and-frame modules is key to making membrane-based capture systems competitive.

2.2 B&W integrated boiler recycle field test

The small pilot test system first used at NCCC was also installed at a B&W Research Center (Barberton, OH) for an integrated field test in 2016 where the CO₂ captured in the sweep step was recirculated to an appropriately-sized coal boiler. While prior computational fluid dynamics modelling and pilot testing of coal combustion with CO₂– enriched air at B&W concluded that CO₂ recycle using sweep membranes appeared feasible as a retrofit, the small pilot field test at B&W would ultimately validate the recycle approach. A three-month test campaign with the boiler operating on natural gas, Powder River Basin coal, and an Eastern Bituminous coal confirmed successful integrated operation of membrane and boiler.

During integrated CO₂ recycle operation, stable combustion and boiler flames were observed over the entire boiler oxygen windbox content range examined (17 – 21%). The air sweep flow rate varied from 500 to 2300 lb/h during the field test and the membrane system was able to achieve capture rates up to 90%. Results of the integrated field test showed minimal impact of CO₂ recycle on combustion pollutant formation (CO, SO2, NOx), no burner modifications are required for stable combustion, and pressure parts modifications would not be needed on existing boilers firing with sweep gas. A modest reduction in boiler efficiency due to recycled CO₂ was measured and found to be consistent with that estimated from modelling and is accounted for in membrane process TEAs.

3.  TCM small pilot field test results

The overall goal of the TCM field test was to validate the transformative potential of scaled-up Gen-2 Polaris membranes and advanced plate-and-frame modules packaged in a containerized form for commercial use. This field test was an important step that demonstrated the performance of these advanced membrane modules at a 10 TPD scale. Specific field test objectives were to identify optimum conditions for different CO₂ capture rates (50 – 90%) through parametric testing, determine system performance under different inlet CO₂ concentrations, and demonstrate low pressure drop of the new planar modules. Through parametric testing, a relationship between the system CO₂ capture rate and CO₂ purity was established under different process operation (i.e. sweep versus non-sweep). By operating the system with inlet CO₂ concentrations up to ~25%, MTR was able to measure system performance under conditions relevant to CO₂ capture from large industrial point sources, such as cement or steel plants. Results from the TCM field test will lead to further refinement of the MTR CO₂ capture process for power plants and other large point sources.

Figure 6(a) shows the MTR small pilot system at the TCM Site for Emerging Technologies, adjacent to the TCM amine plant, while Figure 6(b) is a simplified process flow diagram of the test system. A slipstream of flue gas from Equinor’s Mongstad Refinery residual fluid catalytic cracker (RFCC) unit is sent to the membrane system. After passing through a feed blower, the flue gas goes to the first membrane stage where a vacuum on the permeate side is used to remove CO₂. The membrane permeate is then sent to a second-stage membrane unit. Some of this purified CO₂ can be routed through a spillback line to the front of the membrane system to increase the concentration of CO₂ in the feed from 13% to ~25%. In this way, the feed to the membrane system will mimic the fully integrated case without having to recycle CO₂ to a boiler. The partially treated flue gas that leaves the first membrane step is then sent to the sweep membrane unit. Air flows on the permeate-side of these membranes and removes additional CO₂ from the flue gas. The CO₂-enriched air would be sent to the combustion process in integrated operation, but here it is combined with the other flue gas streams and returned to the TCM infrastructure for local stack venting after analysis. Finally, the cleaned flue gas flows to the stack. Overall, 50 – 90% of the CO₂ in the inlet slipstream can be captured, depending on operating parameters.

Various TCM groups supported the MTR field test throughout the installation, commissioning, operation, and decommissioning phases of the campaign. The MTR test system arrived at TCM in Spring 2021, and MTR personnel were on-site to coordinate execution of installation and commissioning tasks. The test system was commissioned on flue gas in late July and accumulated over 2,200 hours of flue gas operation during the field test. An MTR engineer was on-site for the entire test campaign to operate the test system and coordinate any activities with TCM. The test campaign concluded on March 1, 2022 and the decommissioned test system was completely removed from TCM by mid-June.

During the test campaign, parametric testing of system process variables was conducted to identify optimum conditions for CO₂ capture rates ranging from 50 – 90%. By operating the system under different process modes the inlet CO₂ concentration was varied up to ~25%, which allowed MTR to measure system performance under conditions relevant to CO₂ capture from large industrial point sources, such as cement or steel plants. Through these parametric tests, a relationship between the test system CO₂ capture rate and CO₂ purity was established under different process operation. Figure 7 shows the influence of the inlet flue gas flow rate on the test system performance. Over the flowrate range explored, the overall CO₂ capture rate varied between 61 and 91%, with higher flowrates producing a lower amount of CO₂ capture. This is consistent with expected behavior for a system with a fixed amount of membrane area. The CO₂ purity produced by the 2^nd stage increases from about 86 to 92 mol% (on a dry basis) as the feed flowrate increases. This higher purity is also expected because the higher feed flow rate generates a higher CO₂ partial pressure on the feed-side of the membrane. Overall, the tradeoff in CO₂ purity versus recovery illustrated in Figure 7 is expected behavior for a membrane system.

In addition to CO₂ recovery and purity, another key performance metric for a membrane capture system is the pressure-drop through the modules. During parametric testing, the module stacks in the MTR test system experienced a range of flow rates for which pressure-drops were measured. Figure 8 shows the feed-to-residue pressure-drop for the three module units (Stage 1, Stage 2, and Step 2) as a function of the feed flow rate divided by cross-sectional area (superficial velocity). The pressure-drop of all three module stacks falls on the same curve. This result indicates that the membrane modules performed as expected and there was no evidence of flue gas channeling or flow distribution problems on the feed-side. New 2^nd Stage modules installed in January 2022 that contained a different feed spacer configuration had even lower pressure drop values under the same field test conditions. The comparison of the 2^nd Stage feed-to-residue pressure drop with different feed spacer configurations is shown in Figure 9. Importantly, the feed-to-residue pressure-drop of all membrane stacks is significantly lower than the pressure drop value used in previous TEAs (10.3 kPa or 1.5 psi).

4.  Techno-economic Analysis

Previously, a number of TEAs of the MTR membrane post-combustion capture process have been conducted without input or oversight from TCM. For example, a DOE TEA report (The Pathways Study) on future technologies for post-combustion carbon capture compares the MTR membrane approach favorably with various amine processes [11]. This study shows a membrane system using advanced Polaris membranes and efficient compression equipment

capturing 90% of the CO₂ from an ultra-supercritical coal plant with a 35% increase in cost of electricity (COE) and a cost of CO₂ avoided of <$40/tonne.

In a recent DOE-funded program, MTR completed a AACE Level 2 capital cost estimate with an accuracy of ±15% to demonstrate the economic feasibility of constructing and operating a MTR CO₂ capture system at an existing power plant site. The project evaluated the retrofit of MTR’s post-combustion CO₂ capture technology to the Basin Electric Power Cooperative (BEPC) Dry Fork Station (DFS) Unit 1, located outside of Gillette, WY. The CO₂ capture system was designed to treat the entire flue gas flow from DFS, capture approximately 70% of DFS’s current CO₂ emissions, and pump the high-purity, high-pressure CO₂ to the Carbon Storage Assurance Facility Enterprise (CarbonSAFE) storage complex for geological sequestration. The result of this study determined a cost of capture for the membrane capture system at DFS of $57/tonne CO₂ (in Spring 2022 dollars, which includes >30% inflation compared to Dec 2018 dollars used as a DOE reference).

5.  Conclusions

In summary, MTR is continuing to develop a cost-effective CO₂ capture process using advanced membrane and modules. The TCM small pilot field test validated the potential of the Gen-2 Polaris membrane and advanced planar stack modules. The successful field test moved the Gen-2 Polaris membrane post-combustion capture technology from TRL-5 to TRL-6. In addition to this primary accomplishment, the following key results were achieved:

• The MTR test system designed, built, installed at TCM in Spring 2021, and commissioned on flue gas in July 2021.
• Overall CO₂ capture rates up to 91% and a 2^nd Stage CO₂ purity up to 92 mol% (dry basis) were achieved during parametric testing.
• For all test conditions, the feed-to-residue or sweep side pressure drops were well below the TEA value of 1.5 psi.
• The TCM test campaign ended on March 1, 2022. During the campaign, the MTR test system logged over 2,200 hours of flue gas operation and ran well under summer, fall, and winter conditions.

The Gen-2 Polaris membrane performance and planar module pressure drop data measured during the TCM small pilot field test will be used to help design future MTR post-combustion CO₂ capture systems. One current project that has incorporated experimental data from the TCM field test is a large pilot field test that will capture 150 tonnes of CO₂ per day at the Wyoming Integrated Test Center (WITC) in Gillette, WY during operation in 2023-24. Completion of this large pilot project will advance the MTR post-combustion capture technology to TRL-7 by the mid-2020s and set the stage for future commercial-scale demonstration projects.

Acknowledgements

The material presented here is based on work supported by the U.S. Department of Energy under Award Numbers DE-NT0043085, DE-NT0005312, DE-FE0005795, DE-FE0007553, DE-FE0026414, DE-FE0031591, DE-

FE0031846, and DE-FE0031587. MTR acknowledges the support from TCM during the design, installation, commissioning and decommissioning phases of the field test, assistance in navigating COVID-related international travel bans, and project financial support through in-kind cost share.

Disclaimer: This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that is use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

Demo-scale testing of a hybrid membrane-sorbent system for post- combustion CO2 capture

Dave Gribble^a, Jerrod Hohman^a, Ambal Jayaraman^a, Thomas Hofmann^b, Jay Kniep^b, Tim Merkel^b, Erik Westling^b, and Gökhan Alptekin^a,*

^aTDA Research, Inc. Wheat Ridge, CO 80033 USA

^bMembrane Technology and Research, Inc., Newark, CA 94560 USA

Abstract

TDA Research is developing a novel hybrid membrane-sorbent system for the post-combustion capture of carbon dioxide (CO₂) from supercritical pulverized coal (PC) power generation facilities or large point industrial emitters. The novel design incorporates a 1^st-stage continuous membrane separator developed by Membrane Technology and Research, Inc. (MTR) with a dual-bed radial-flow sorbent contactor (designed and developed at TDA). Testing of a pilot unit at 1 MW_e scale has been conducted at the site for emerging technologies at the Technology Centre Mongstad (TCM) in Mongstad, Norway using residue fluid catalytic cracker (RFCC) flue gas from Equinor’s Mongstad refinery.

A 10-month field demonstration concluded in March 2022—marking the culmination of a 4-year R&D program funded by the US Department of Energy’s National Energy Technology Laboratory with a significant cost-share provided by TCM and the TCM ownership group, a joint venture between Gassnova (on behalf of the Norwegian state), Equinor, Shell, and TotalEnergies. The total operational time on flue gas was 4,001 hours. During that period, the system demonstrated CO₂ capture at or above 90% for 1,789 hours (cumulative) of which 309 hours were at or above 95%. In total, the hybrid membrane-sorbent system processed 1,890 MT of CO₂ at an average capture efficiency of 87.1%—including parametric testing and process upsets.

Prior to the 1 MW_e field test, TDA and MTR developed a techno-economic analysis with the assistance of Dr. Ashok Rao and the Advanced Power and Energy Program (APEP) at the University of California, Irvine (UCI). The TEA utilized performance data from earlier tests of the system components along with integrated test results obtained at a smaller scale. The analysis indicates that the hybrid membrane-sorbent CO₂ capture system (integrated with a supercritical PC plant) achieves a net plant efficiency of 29.45% (HHV basis)— higher than that achieved by amine- based CO₂ capture system (28.40%; HHV basis). The cost of CO₂ captured (excluding TS&M costs) for the hybrid membrane-sorbent system is estimated at $43.30/tonne which is well below the $56.49/tonne cost for amine-based system (2011 $ basis). The TEA analysis and Aspen Plus^® model were developed at APEP without input or oversight from TCM or the TCM ownership group. An updated TEA that reflects the results from the TCM field demonstration and utilizes the latest DOE Revision 4 basis is currently underway.

Keywords: CO₂ capture; post-combustion; adsorbent; membrane-sorbent hybrid, point source emissions; power plant

1.       Introduction

TDA recently completed a successful engineering-scale field demonstration of a radial flow CO₂ contactor. The demonstration was performed at the world’s largest CO₂ capture test center, the Technology Centre Mongstad (TCM) in Mongstad, Norway. TCM is majority-owned by the Norwegian state through Gassnova (roughly 74%) with industrial partners Equinor, Shell, and TotalEnergies holding a combined stake of roughly 26%. Testing was primarily performed using flue gas from the Residue Fluid Catalytic Cracker (RFCC) at the adjacent Mongstad Refinery (owned and operated by Equinor). A short test was also performed at the end of the campaign using flue gas from the Combined Heat and Power (CHP) facility. The RFCC flue gas has a CO₂ concentration of roughly 14% by volume (a close proxy for flue gas from coal-derived power generation facilities), while the CHP unit provides a lower CO₂ concentration (roughly 4% by volume) which is more representative of Natural Gas Combined Cycle (NGCC) flue gases. The field demonstration was part of a larger 4-year program operated with funding from the U.S. government via the Department of Energy’s National Energy Technology Laboratory (NETL) under contract DE-FE-0031603.

In the first year of the project, TDA, Membrane Technology and Research, Inc. (MTR), and TCM collaborated on the design and engineering in order to ensure a successful demonstration that would integrate into the TCM facility while combining an existing membrane system with TDA’s sorbent contactor. Concurrent with the system design, TCM finalized development plans for their emerging technologies site (an area designated for large demonstrations of novel technologies) in order to provide all the requirements for a successful campaign. The design work concluded with a 3-party Process Hazard Analysis (PHA) facilitated at TDA. Following the PHA, the project transitioned into system fabrication and shipping (year 2). During the 2^nd year of the project, MTR completed all of the required membrane system modifications. TDA fabricated all of the new equipment required for the radial-bed contactor, and TCM completed their upgrades in order to host the technology demonstration. During this phase MTR also fabricated new Gen-1 Polaris™ membrane modules for the demonstration and TDA prepared sufficient sorbent to fill the radial beds 2 times each. At the end of year 2, the equipment was shipped to TCM in Norway (Figure 1).

On-site commissioning for the nominal 1.0 MW_e (roughly 8.6 million MT CO₂/year) demonstration began in May of 2021 and testing with RFCC flue-gas commenced on 16 June 2021. The test campaign ran through 18 March 2022 as a single 10-month campaign wherein the hybrid membrane-sorbent system was operational (processing flue gas) for a total of 4,001 hours. Excluding planned maintenance and holiday stoppages, the downtime during the campaign totaled 969 hours—equating to an overall system availability of roughly 80.5%. The system was considered to be operational only when processing flue gas (RFCC or CHP). Therefore, all of the cumulative stoppages, system restarts, and periods where the membrane skid was processing ambient air are included in the offline total. This also included plant stoppages and RFCC unit upsets that occurred upstream of the TDA/TCM battery limits.

2.       Process Description

The hybrid system is a 2-stage system that incorporates a membrane contactor followed by a sorbent contactor; Figure 2 provides a simplified process block diagram. The MTR membrane is a solution-diffusion polymeric composite membrane, and the sorbent is a mesoporous carbon material. For the hybrid system, TDA also developed a new dual-bed radial contactor. The new design was based on data from a previous (roughly 1/20^th-scale) experimental

systems. The scaled-up contactor incorporated a mechanism for selectively blocking a portion of the sorbent bed so that we could test the modularity of the sorbent contactor.

This CO₂ capture process was designed to tightly integrate the post-combustion capture system with a supercritical pulverized coal (PC) facility. The commercial system integration was designed to maximize the effectiveness of both the membrane and the sorbent sub systems. In order to achieve this optimized system design, the 1^st-stage membrane performs the CO₂ product recovery function by stripping roughly half of the incoming CO₂ from the flue gas stream while passing the remainder (the retentate stream) to the sorbent contactor. The sorbent contactor then strips the remainder of the CO₂ from the membrane’s retentate—producing a CO₂-depleted flue gas exhaust that can be released directly to the atmosphere. The absorbed CO₂ is then removed (the sorbent regenerated) by a concentration-swing process using atmospheric air which in turn produces a CO₂-enriched air stream. The CO₂-enriched air is recycled to the coal boiler where it serves as the combustion air—resulting in a CO₂-enriched flue gas. The resulting flue gas (due to the increased CO₂ content) allows for a larger CO₂ partial-pressure difference across the 1^st-stage membrane which in turn results in increased CO₂ transport and improved CO₂ product purity (membrane permeate). The 2-bed sorbent contactor improves the CO₂ capture efficiency of the carbon capture system (95–98% capture efficiency can be achieved as desired for future deep decarbonization applications).

3.       Mechanical Stability of the Sorbent

TDA prepared its proprietary carbon sorbent material at our facility in Golden, Colorado, USA and shipped the material to TCM in a series of steel drums. Each drum was filled at TDA’s facility and the tare mass (drum) and net mass (sorbent) was recorded and noted on the individual drums prior to shipment. Each drum contained roughly 100 kg of sorbent material packaged in a heavy-duty polymer drum liner. In order to limit the possibility of physical attrition during transit, the sorbent beds were shipped empty and were filled with sorbent onsite using a manual tilting drum handler (Figure 3) attached to chain hoists on small jib cranes (1 per vessel). The jib cranes and chain hoists were also used for removing and installing the full-access sorbent vessels lids. This allowed the entire filling operation to be performed by a small crew without relying on heavy equipment.

A total of 9 drums were required for each vessel. TDA used a cordless concrete vibrator (18V Makita model XRV02Z with 8’ wand and 13,000 vibration/minute motor) to pre-settle the material. The vibrator was inserted into the sorbent material vertically and the material vibrated in order to compact the bed. This was repeated around the entire circular area of the bed after each drum was added through the 7^th drum. Thereafter, the material was manually raked/brushed using various hand tools in order to uniformly fill the entire bed. The vibrator could not be used effectively when the bed level approached the top of the basket or the vibration would have caused material to be ejected from the bed. The beds did not accommodate the entirety of their respective 9^th drums, so the remaining mass was recorded using a digital scale attached between the chain hoist and the drum handler. The total sorbent mass for each vessel is listed in Table 1 (T-810) and Table 2 (T-820).

During the loading process, almost no dust formation was observed. After the initial sorbent loading the sorbent beds were de-dusted under flowing air. The regeneration air blower was first used to de-dust each bed individually

unidirectionally. Following the initial de-dusting, the blower was stopped and fines were collected from the in-line filter receiver (> 98% removal for particle sizes of 3 μm and larger). The regeneration air blower was used to further de-dust the beds while switching the flow between beds with additional fines collection afterward. The membrane skid was then started and forced to run air to the sorbent system so that the beds could be further de-dusted in an alternating, bi-directional manner. During the de-dusting operation and periodically during the first month of operation, fines were collected from the filter receiver. The cumulate mass of fines collected (from both beds) is shown in Figure 4.

Following the de-dusting, both vessels were opened, the sorbent levels inspected, and additional sorbent was added to each vessel in order to fill the vessels as completely as possible. The amount of material added is listed in Table 1 and Table 2. Because the total amount of collected fines was very small, we attribute all of the added material to settling which was minimal (4.5% and 2.0% for vessels T-810 and T-820 respectively.

When inspecting the sorbent material, the settling was appeared to be non-uniform with the largest voids (generally measuring less than 1–2” deep and covering only a few square inches per void) being located along the inner and/or outer walls and primarily in the vicinity of the spoke-like structures that connected the inner and outer sorbent baskets. It is speculated that the care taken to avoid spilling material into the outer gas distribution channel and the effort to avoid sorbent debris accumulation on the top of the spoke-like frame where the inner basket gaskets are installed could be the cause of the small voids along the top of each surface which were exacerbated during the de-dusting campaign. These same void locations were shown to develop in future inspections as well, but never grew to significant size (depth or area), and at no point appeared to generate open pathways across the beds (in the radial direction) which could lead to channelling and allow the flue gas to bypass the sorbent.

Figure 5 shows a series of images across the top of sorbent vessel T-810. These images were recorded during the final sorbent inspection and indicate the amount of settling observed between the intermediate inspection/top-off (September 2021) and the end of the field test (March 2022). The maximum settling is observed along the outer sorbent retention basket (leftmost image) and measures a maximum depth of roughly ¾” (see the insert; Figure 5 for scale reference). Scanning across the middle of the bed (radial direction) shows almost no settling.

The mesoporous carbon sorbent material performed well physically throughout the testing with only minimal fines being collected within the in-line filter receiver. Based solely on physical changes (attrition or fines generation), the sorbent lifetime is expected to far exceed the 3-year target lifetime assumed at the onset of the campaign.

4.       Radial Bed Pressure Drop

A key feature of the radial sorbent contactor is the low pressure drop that can be achieved by distributing the flow over a large surface area while minimizing the bed depth. The flow enters the bed from the center, is distributed within the cylindrical inner distribution channel, travels radially outward through the sorbent material, and recombines in the outer distribution channel which comprises the luminal space between the outer sorbent basket and the inner vessel

shell. In the case of the regeneration air, the flow pathway is reversed as the air is swept from outside to inside across the sorbent.

Prior to fabricating the sorbent vessels, TDA performed theoretical pressure drop calculations based on a cylindrical form of the Ergun equation and compared the results to CFD modeling performed by GTI Energy using ANSYS Fluent. The experimental pressure drop data was found to be in good agreement with the theoretical predictions as shown in Figure 6. With a flue gas flow rate ranging from 2,000–2,500 kg/hr, the total ΔP through the radial sorbent beds was measured at ≤ 20 mbar. As a comparison, the membrane unit, treating roughly the same flue gas flow and rejecting a similar amount of CO₂, generated roughly 120 mbar of pressure drop. TDA observed that the pressure drop through the sorbent contactor remained constant for a given flow rate throughout the test campaign—another indication that the sorbent is not attritting or otherwise physically breaking down as a result of prolonged flue gas exposure and/or flow/pressure cycling.

5.       CO2 Capture Efficiency

In this demonstration, TDA calculated the CO₂ capture efficiency for the hybrid membrane-sorbent system as the single-pass CO₂ removal for the CO₂ present in the incoming flue gas stream using Equation 1.

Using the CO₂ flow rates from the June 2021 example (Figure 7), the calculated CO₂ capture efficiency is (391 kg/hr – 18 kg/hr) / 391 kg/hr * 100 = 95.4%.

Throughout the test campaign, the system processed roughly 1,890 MT of CO₂ with an overall capture efficiency of roughly 87.1%. These totals include several parametric studies during which the system was intentionally operated at sub-optimal conditions in order to characterize the impact of various operating parameters. When operating under

preferred conditions, the system demonstrated high-efficiency capture—including 1,789 hours of cumulative operation with a capture efficiency at or above 90%. Within that subset of the demonstration, so-called deep decarbonization was also demonstrated. In total, the system ran for 307 hours with capture efficiencies at or above 95%—showing that the membrane-sorbent system can achieve CO₂ removal efficiencies of 95% and above (up to a maximum sustained CO₂ removal efficiency of roughly 98%).

Throughout the field test campaign, the sorbent beds were cycled more than 160,000 times (some valves operated twice per cycle—in excess of 320,00 times during the campaign). Despite the high cycle count, the pneumatic butterfly valves used for the main process lines performed well—requiring no valve or actuator maintenance. However, a few of the valve position switches did require re-alignment after the initial system startup.

In a series of parametric testing, we determined that the absorption cycle time has a significant impact on the capture efficiency with shorter cycle times (down to 25 seconds/bed, the minimum tested) showing increased capture efficiency and improved hourly sorbent loading. We identified that a 40-second per bed absorption cycle was generally adequate to reach the 90% capture design target.

We also determined that the regeneration air flow rate has a significant impact on the CO₂ capture efficiency while the regeneration air temperature has a lower impact. In the case of the regeneration air flow rate, we constricted the parametric testing to flow rates ranging from 70–110% of the retentate flow rate (mass basis). This constrained regeneration air flow rate is in good agreement with what is expected in the power plant operation (i.e., since the coal boiler combustion air is used as the regeneration gas, the flow rate cannot exceed the rate required for coal combustion). Similarly, the regeneration air temperature range was constrained to modest temperatures ranging from 30–60°C.

Figure 7 shows the CO₂ capture efficiency (top chart) and the mass flow rates for CO₂ in the various streams (lower chart) for a 48-hour deep decarbonization test run from 3–5 August 2021 (roughly 38 days into the test campaign). The sorbent material was not replaced or topped-off/modified prior to this test. Data from a separate deep decarbonization test is shown in the form of a simplified process flow diagram (Figure 7) with the average hourly CO₂ flow rates indicated for each process stream. This test from earlier in the campaign (30 June) also shows deep decarbonization potential with a single-pass capture efficiency of roughly 95.4%. When operated under preferred conditions, the hybrid membrane-sorbent system demonstrated deep decarbonization potential with sustained capture efficiencies in excess of 95% and occasional performance in excess of 98% CO₂ capture.

6.       Sorbent Performance Stability

The sorbent material lifetime does not appear to be limited by physical degradation. To assess the impact of potential flue gas contaminants and to monitor for potential long-term changes in the sorbent performance, TDA

utilized a “baseline” test that was repeated periodically throughout the field test campaign. In these tests, the goal was to return to a specific set of operating parameters across both the membrane and sorbent sub-systems in order to compare the net capture efficiency throughout the campaign, thereby demonstrating stable system performance over a prolonged test.

Figure 8 shows a timeline of the capture efficiency obtained from repeating tests under nearly identical conditions at the beginning, middle, and end of the test campaign. The baseline test was performed at 3 different regeneration air

(A) to retentate (R) mass flow ratios (0.8, 0.9, and 1.0). The data shows a small increase in performance for the highest A/R ratio, but is relatively flat at the lower two ratios. In all cases, the deviation falls within the uncertainly, demonstrating highly stable performance over the 10-month test campaign.

During the final sorbent inspection in March 2022, samples were collected from throughout the sorbent beds. These samples are currently being shipped to TDA’s laboratory where they will undergo a thorough physical and chemical analysis to detect any changes that occurred during the course of the test campaign.

Technoeconomic Analysis

In our hybrid process (Figure 9, left), the flue gas from the wet Flue Gas Desulfurization (FGD) scrubber enters the membrane module at 30–55°C. The membrane removes about 50–55% of the CO₂ and almost all of the H₂O from the flue gas. Only a mild vacuum (0.2 bar, 3 psia) is needed to facilitate the CO₂ transport across the membrane because the CO₂ concentration of the gas is increased by CO₂ re-circulation through the steam boiler. The membrane permeate stream consisting of CO₂/H₂O is compressed and cooled and the water is condensed so that the CO₂ can be recovered as a high-purity product. CO₂ purity on a dry basis is expected to be between 65–85%, further purification to bring the CO₂ purity to 95% (dry basis) and reduce the O₂ concentration down to below 10 ppm is accomplished using a cryo- cooler/distillation step. The retentate from the membrane is treated by TDA’s 2-bed sorbent system, which removes additional CO₂ to achieve the 90% CO₂ recovery.

When the sorbent is saturated with CO₂, it is regenerated using the boiler combustion air. This ambient air stream has trace CO₂ (c.a. 400 ppmv) and performs well as a sweep gas for sorbent regeneration without requiring the additional energy input needed for a temperature and/or pressure swing. The CO₂-laden air (regeneration stream) from the sorbent bed is fed into the coal boiler as a recycle—resulting in a CO₂-enriched boiler exhaust that is critical for effective operation of the membrane.

Based on a 550 MW_e power plant, we designed the sorbent system for the hybrid CO₂ capture process as 4 independent 4-bed trains with each train handling roughly 137.5 MW_e of flue gas. The rough foot print of the system (just the beds) is approximately 19.3 m x 15.5 m (63’1” x 50’9”) with a total area of just under 300 m² (3,230 ft²) with a maximum height of 14.9 m (48’11”). A 3D layout of the sorbent sub-system for TDA’s hybrid membrane-sorbent CO₂ capture process is shown in Figure 9 (right).

All of the mass and energy balance calculations used in the technoeconomic analysis were developed using Aspen Plus^® simulation software integrating the membrane and sorbent separators into the overall power plant. The basis for this design is identical to that used in the Case 12 of the DOE Baseline Study (Rev. 2a basis) [1] consisting of an amine-based CO₂ capture technology. The hybrid system was modeled to provide the same net power of 550 MW_e as the DOE Case 12. The net plant efficiency (HHV basis) for the TDA hybrid membrane sorbent process integrated with a supercritical power plant is calculated to be 29.45%, which is 3.7% higher than the DOE Case 12 [1]. In comparison to the DOE case 12 [1], the raw water withdrawal for the hybrid membrane-sorbent system is reduced by roughly 11%. This reduction is primarily due to using solid sorbent and membranes for CO₂ capture instead of aqueous amines.

The total plant cost (TPC) for the 550 MW_e net supercritical power plant equipped with TDA’s hybrid membrane- sorbent process is estimated at $3,006/kW on $2011 basis, which is roughly 15.6% lower than the DOE Case 12 cost of $3,563/kW [2]. The PC boiler cost constitutes about 25% of the total plant cost for Case 3 while flue gas cleanup constitutes about 14%, and both the CO₂ removal system and steam generator along with its auxiliaries constitute about 11% each.

The cost of electricity (COE) is then calculated using a capital charge factor (CCF) of 0.1243 and a capacity factor (CF) of 85%, while using $10.00/tonne of CO₂ ($9.45/MWh) as the transportation, storage, and monitoring (TS&M) costs per the Rev. 2a basis DOE baseline study [2]. The 1^st year levelized cost of electricity without the TS&M costs for the hybrid process integrated power plant is $121.9/MWh, which is roughly an 11.3% reduction of the DOE Case 12 cost. The 1^st year levelized CO₂ capture cost without the TS&M costs is $43.30/tonne—a 24.4% (or $13/tonne) reduction from the DOE Case 12 cost. The 1^st year levelized cost of electricity inclusive of the TS&M costs is at

$131.3/MWh—roughly a 10.9% reduction compared to the DOE Case 12 cost. The 1^st year levelized CO₂ avoided cost inclusive of the TS&M costs for the hybrid process is at $73/tonne, which is roughly $23/tonne lower than the DOE Case 12 cost.

Acknowledgements

This material is based upon work supported by the Department of Energy National Energy Technology Laboratory under Award Number DE-FE0031603. TDA also thanks the Technology Center Mongstad for providing significant pre-demonstration design support, hosting the demonstration, commissioning and decommissioning assistance, navigating COVID-related travel bans, and for providing financial support through in-kind cost share. TDA also thanks our collaborators MTR, GTI Energy, and UCI who contributed to the success of this project.

Disclaimer

This report was prepared as an account of work sponsored by an agency of the United States Government.

Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

References

• National Energy Technology Laboratory. Cost and performance baseline for fossil energy plants, Volume 1: Bituminous coal and natural gas to electricity, Revision 2a. National Energy Technology Laboratory; September 2013. DOE/NETL-2010/1397. https://netl.doe.gov/projects/files/CostandPerformanceBaselineforFossilEnergyPlantsVol1BitCoalandNGtoElecRev2aSept2013_090113.pdf.
• National Energy Technology Laboratory. Updated costs (June 2011 basis) for selected bituminous baseline cases. National Energy Technology Laboratory; August 2012. DOE/NETL-341/082312. https://netl.doe.gov/projects/files/UpdatedCostsJune2011BasisforSelectedBituminousBaselineCases_082312.pdf.